Urgent.News

What's breaking now, across thousands of outlets.

Tech

Python में sliding window से ट्रांसक्रिप्ट में हाइलाइट क्लिप्स कैसे ढूँढें

अगर ट्रांसक्रिप्ट में हर शब्द का टाइमस्टैम्प है, तो लंबी रिकॉर्डिंग में छोटी क्लिप्स के कैंडिडेट ढूँढने के लिए किसी मॉडल की ज़रूरत नहीं है। तय लंबाई की एक विंडो हर शब्द से शुरू करें और उसे इस हिसाब से स्कोर दें कि उसमें कितने शब्द आते हैं। इससे वे हिस्से मिल जाते हैं जहाँ सबसे ज़्यादा बात कही गई है। osFoundry डेस्कटॉप ऐप का वीडियो एडिटर ठीक यही चीज़ अपने Suggest clips बटन में देता है, Rust में (…

This article explains how to find the best clips in a transcript using a sliding window approach in Python. The method involves creating candidates by starting a window at the beginning of each word and scoring them based on the number of words they contain. The algorithm then selects the best candidates based on their scores, which can be adjusted by changing the window size.

The example provided uses an interview with astronaut Christina Koch and her Whisper transcript, which contains 306.97 seconds of audio and 1,007 words.

Brief written by urgent.news from Dev.to's own syndicated text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in Tech

Large Image Uploads: A Node.js Sharp and Multipart Retention Runbook

Short answer: for a gaming platform keeping per-tenant backups and restoring a selected snapshot, let deletion be a reviewed policy decision, and let a completed multipart object be the only input to…

  • Deletion is deliberate policy decision, not automatic
  • Completed multipart uploads feed Node.js Sharp pipeline
  • Originals private, derivatives stored separately

RTX 3090, RTX 5090, A100 and H100 on vLLM: the cheapest GPU flips with your answer time

I measured four GPUs serving the same 7B model on vLLM. Which one is cheapest per token has no single answer. It flips with how fast your answers must be.

  • H100 is cheapest for answers under 2 seconds at $0.441 per 1M tokens
  • RTX 5090 becomes most cost-effective for answers up to 8 seconds at $0.050
  • H100 remains most affordable with max requests (64) at $0.162

Tabs in One: соберите все вкладки Chrome в один столбец и освободите оперативную память

Если вы это читаете, у вас прямо сейчас открыто больше 40 вкладок. Chrome съел 4 ГБ памяти, вентилятор ноутбука воет, а вкладку, которую вы открыли 20 минут назад, уже не найти.

  • Tabs in One consolidates all open Chrome tabs into a single column
  • One-click restore of individual or all tabs without loss of data
  • Automatic backups and export/import features for tab management

More from Saturday 7 November →